In a simple conventional capacitive sensor design, the common electrode, which is mobile, forms part of an armature resiliently held between the two fixed electrodes. In this case, the capacitive sensor may be capable of performing a measurement along one direction of movement of the moving electrode. At rest, this common moving electrode is normally equidistant from the two fixed electrodes in order to have two capacitors with an equal capacitance value. The common moving electrode can move some distance in the direction of one or other of the fixed electrodes under the action of a force. Thus the capacitive value of each capacitor varies inversely. The electronic interface circuit connected to the capacitive sensor thus enables an analogue output signal to be supplied. This analogue output signal takes the form of a voltage dependent on the capacitance variation of the two capacitors.
This electronic interface circuit for a capacitive sensor is disclosed in the article by Messrs H. Leuthold and F. Rudolph, which appeared in the journal entitled “Sensors and actuators” A21-A23 (1990), pages 278 to 281.
The capacitive sensor may be an accelerometer for performing an acceleration measurement in conjunction with an electronic interface circuit. It may be a single axis accelerometer like the aforementioned capacitive sensor, or a multi-axis or tri-axis accelerometer for performing a measurement in three directions X, Y and Z. A tri-axis MEMS accelerometer of this type may include a single mass, i.e. a common inertial mass for the three pairs of differential capacitors, or three masses for the pairs of capacitors. In the first case, a single common electrode and six fixed electrodes are provided, whereas in the second case, one common electrode and two fixed electrodes are provided for each pair of capacitors.
Both the electronic circuit and the capacitive sensor, such as a MEMS accelerometer, are made in a semiconductor substrate. Consequently, stray capacitances at the electronic circuit input are added to the capacitances of the capacitors of the MEMS capacitive sensor. These stray capacitances do not depend on the motion of the moving electrode, which consequently creates non-linearities and also lowers the sensitivity or gain of the electronic circuit. The same is true with the MEMS capacitive sensor, where the potential of the substrate during operation of the sensor also creates non-linearities. Thus, the mean electrostatic force is not zero in the sensor and electronic circuit off mode. Because of the influence of the substrate potential on the electrostatic force, this leads to a variation in the measured real force, which is applied across the common moving electrode, which is a drawback.
To carry out a force, acceleration or pressure measurement using the electronic circuit, the fixed electrodes of two capacitors or pairs of capacitors are biased or excited cyclically by voltages of opposite polarity relative to an off reference voltage. By biasing or polarizing the two fixed electrodes at different voltage levels, the charge difference across the moving electrode can be measured and converted into at least one electronic circuit output voltage. When the output voltage or voltages are stabilised at their final value, the total charge across the moving electrode becomes zero. Consequently, these output voltages are supplied sampled to a processing circuit.
Since the measurement of a force, acceleration or pressure is dependent on the aforementioned non-linearities and on any voltage offset linked to unmatched electronic components, EP Patent No. 1 835 263 B1 proposes a solution to this problem. A symmetrical double structure is proposed in the electronic circuit, particularly with two integrators following the charge transfer amplifier. Each integrator supplies a corresponding analogue output voltage at output according to positive electrode biasing or negative electrode biasing. Because of this, a voltage offset due to technology or to the variation in supply voltage can be minimised or eliminated using the two analogue integrator output voltages. Moreover, the substrate potential is no longer of any importance given that the electronic circuit includes an identical double structure operating in total symmetry.
However, one drawback of the electronic circuit of EP Patent No 1 835 263 B1 is that it supplies output signals, such as output voltages, in analogue form. This requires the use of two integrators. This means that it is not possible to sufficiently reduce the size of the integrated components and the electrical power consumption of the integrated electronic circuit. Moreover, the electronic circuit is only arranged to perform a measurement with one capacitive sensor along a single measurement axis.
It is thus preferred to make an electronic circuit which supplies digital measuring signals at output. WO Patent Application No. 2004/113930 A1, which discloses an electronic circuit of this type connected to a single axis or multi-axis capacitive sensor for measuring an acceleration, can be cited in this regard. A logic circuit specific to each measurement axis, which processes digital measuring signals, is provided after the charge transfer amplifier, which is connected to the common moving electrode. The output of each logic circuit supplies a binary measuring signal representative of a measuring voltage level dependent on the movement of the moving electrode relative to the fixed electrodes for each axis in succession. The binary measuring signal for each axis is supplied in succession to a digital-analogue converter. In one phase of each measuring cycle for a selected axis, this converter supplies a measuring voltage to the electrodes alternately with a phase of biasing the fixed electrodes at a high voltage and a low voltage from a supply voltage source. The binary signal obtained at the output of each logic is incremented or decremented by one unit at each series of measuring phases, until the total charge across the moving electrode becomes zero. Although the size of the electronic components and the electrical power consumption are reduced, the aforementioned non-linearities and voltage offsets are not removed, which is a drawback. Moreover, the stabilising time of the digital output signal for each measurement axis is relatively long, which is another drawback.
Like the preceding document, WO Patent Application No. 2008/10773 A1 discloses an electronic interface circuit for an acceleration measuring sensor. An analogue measurement signal is stored after a charge transfer amplifier in one phase of a measuring cycle after the fixed electrodes of the capacitor have been biased. The analogue signal is converted into a digital signal stored in a logic of the electronic circuit. The stored digital signal is subsequently converted by a digital-analogue converter into an analogue return signal in the form of a voltage, which is applied to all the sensor electrodes in a successive phase of each measuring cycle. In a measuring cycle, the fixed electrodes are biased in succession by a first biasing and a second biasing which is the reverse of the first biasing. This enables leakage currents to be removed from the electronic circuit. However, a large number of steps are necessary to obtain a physical parameter measuring signal at output, which is a drawback.
EP Patent Application No. 2 343 507 A1 discloses an electronic interface circuit for a single axis or tri-axis measuring sensor. The measuring signals are digitally processed after the charge transfer amplifier in a logic unit. Following positive biasing and negative biasing, digital measuring signals are stored in corresponding registers of the logic unit. A digital-analogue converter is also used for successively converting the digital signals for each axis in a measuring cycle into a voltage at the sensor electrodes. To obtain final measurement values for the acceleration for each axis, a dichotomy algorithm is first of all used in the logic unit for a certain number of measuring cycles, prior to finishing with oversampling steps. With this dichotomy algorithm, the measurement always starts, during each conversion, at half the measurement range, in particular at VREG/2. If an error occurs during this first measurement with a large change step in the logic unit, the final value at the end of all the measuring cycles will inevitably be erroneous, which is a drawback.